Skip to content
LeitaguDocs

Loading…

Pages

Platform

Reading this hub as an AI agent

The machine-readable ways into this hub (Markdown pages, llms.txt, one-file export and a JSON index) and how far its statements can be trusted.

Updated 16 October 2026View as Markdown

Every page of this hub is also available as plain Markdown, and the whole hub can be fetched as one file or read through a JSON index. Statements about how Leitagu behaves are taken from the code and the app READMEs in the repository, and anything planned is marked as planned. Each page's front matter says when it was last updated.

Addresses

Address What it returns
A page's address plus .md, for example /product/why.md, or /index.md for the start page The page's Markdown, front matter included
/llms.txt The hub's index in the llms.txt format: a short summary, then one link per page, grouped by section, pointing at the .md addresses, each with the page's description
/llms-full.txt Every page in reading order in one Markdown file. Each page starts with a short header giving its title, address, section and update date.
/index.json A machine-readable index: for each page its path, address, Markdown address, title, description, section, update date and headings (level, text and anchor id)

Each HTML page also points to its Markdown version with <link rel="alternate" type="text/markdown">. Links between pages in the Markdown are relative .md links, so they resolve against the .md addresses.

Front matter

Every page starts with:

---
title: Analytics definitions
description: One sentence saying what the page covers.
section: partners
order: 3
updated: 2026-10-16
---
  • section is one of: start, product, shoppers, partners, platform, reference.
  • updated is the date the page was last revised. Use it to judge freshness: a page older than a change in the code may not describe that change.

How far to trust it

  • Sourced from the code. Behaviour is described from the apps in apps/, the shared packages in packages/ and their READMEs. Where the hub and the code disagree, the code is right and the page is out of date.
  • Planned is labelled. Features that are not built say so, usually with "Planned" or "Not built". What is not built yet lists the main gaps.
  • Example data. All 50 products in the shop are example data. No real store has connected yet.
  • Research is a draft. Company facts in the CRM's research dataset come from search results and are unverified until checked in the Business Register.

Where to start